Detection of Novel Genomic Markers for Predicting Prognosis in Hepatocellular Carcinoma Patients by Integrative Analysis of Copy Number Aberrations and Gene Expression Profiles: Results from a Long-Term Follow-Up
Hyo Jung Cho,1 Soon Sun Kim,1 Hee Jeong Wang,2 Bong Wan Kim,2 Hyeseong Cho,3,4 Junghee Jung,5 Samuel Sunghwan Cho,5 Jai Keun Kim,6 Jei Hee Lee,6 Young Bae Kim,7 Min Jae Yang,1 Byung Moo Yoo,1 Kwang Jae Lee,1 Sung Won Cho,1 and Jae Youn Cheong1
Abstract
The aim of this study was to explore novel genomic biomarkers predicting hepatocellular carcinoma (HCC) prognosis by integrative analysis of DNA copy number aberrations (CNAs) and gene expression profiles. Array comparative genomic hybridization and expression array were performed on 45 and 31 HCC samples, respectively. To identify functionally important genes, concordant results of DNA copy number and gene expression were retrieved by integrative analysis. Cox regression analysis indicated that the CNAs in 192 genomic regions were significantly associated with overall survival (OS; p<0.05). Integrative analysis capturing concordant results demonstrated that the low expression of TLE4 (p=0.041) and XPA (p=0.006) was associated with poor OS. In the analysis of tumor recurrence, 514 genomic regions with CNAs were associated with recurrence. Integrative analysis revealed that the overexpression of 16 genes, including FGR (p=0.003), RELA (p=0.049), LTBP3 (p=0.050), and RIN1 (p=0.023), was significantly associated with shorter time to tumor recurrence. On multivariate analysis, FGR and XPA were independent risk factors of early recurrence and poor OS, respectively. Integrated analysis of CNAs and gene expression profiles correlated with long-term follow-up data successfully identified potential prognostic markers predicting survival and tumor recurrence in patients with HCC who underwent surgical resection.
Introduction
Hepatocellular carcinoma (HCC) is the fifth most upregulation of the gene (Hyman et al., 2002; Pollack et al., common neoplasm in the world and represents the 2002; Huang et al., 2006). A recent study demonstrated that third leading cause of cancer mortality (Parkin et al., 2001). Prognosis of HCC is known to be poor; however, genomic markers related to patient prognosis have not yet been identified.
Analysis of DNA copy number aberrations (CNAs) is particularly important in cancer research because the amplification of oncogenes and/or deletion of tumor suppressor genes are critically involved in the pathogenesis of cancer development and progression (Kinzler and Vogelstein, 1996). CNAs are likely to influence gene expression; however, they are not essential to the transcriptional expression of most genes in cancer (Stranger et al., 2007). Overexpression of a single-nucleotide polymorphism and copy number variation genotypes were associated with 84% and 18% of the variation in those gene expression traits, respectively (Stranger et al., 2007). Another study showed that 62% of highly amplified genes demonstrated moderately or highly elevated expression in breast cancer (Pollack et al., 2002).
Array comparative genomic hybridization (CGH) has enabled the high-resolution detection of CNAs (Pollack et al., 1999). Combined analysis of array CGH and gene expression data, followed by functional validation, is regarded as a powerful tool for the discovery of cancer driver genes from genome-wide array-based strategies. In HCC, CNAs occur frequently and contribute to gene expression profiles (Katoh et al., 2005). Several studies have reported the results of integrative analysis of CNAs and expression profiles in HCC samples, but few have focused on the prognostic impact of candidate genes in patients with HCC (Midorikawa et al., 2006).
In the current study, we applied integrative analytical methods to correlate CNAs obtained from array CGH and the corresponding gene expression profiles obtained from oligonucleotide microarray in HCC samples. We then searched for the genomic markers predicting patient outcome using long-term follow-up data regarding overall survival (OS) and tumor recurrence in HCC patients who underwent surgical resection.
Materials and Methods
Patient characteristics and HCC sample collection
HCC specimens were obtained from 50 patients who had underwent hepatic resection for HCC treatment at the Ajou University Hospital (Suwon, Republic of Korea) between March 1995 and March 2003. Control liver samples were obtained from 25 patients who had underwent hepatic resection for management of metastatic liver tumor from gastrointestinal malignancies. The biospecimens for this study were provided by the Ajou Human Bio-Resource Bank, a member of the National Biobank of Korea, which is supported by the Ministry of Health and Welfare. Specimens were reviewed by an expert pathologist, and pathologic diagnosis of HCC was based on international consensus panel (2012).
Data about the gender, age, etiology of liver disease, tumor number, tumor size, vascular invasion, alphafetoprotein (AFP), histologic grade of HCC according to the Edmondson–Steiner grading system, HCC stage according to the modified Union for International Cancer Control (UICC), OS, and tumor recurrence were collected from medical records. OS was defined as the period from treatment day to the last follow-up or death caused by HCC. Disease-free survival was defined as the interval between treatment day and the confirmation day of HCC recurrence. The recurrence of HCC was defined as newly developed lesions on CT, which showed early enhancement in the arterial phase with washout in the late phase. The median follow-up period was 83 months (range=3–202 months).
Among the obtained specimens, array CGH analysis and messenger RNA (mRNA) expression profiling were performed in 45 and 31 HCC tissues, respectively. Twenty-six specimens of HCC were analyzed for both array CGH and expression array. The baseline clinical and tumor characteristics of patients are depicted in Supplementary Table S1 (Supplementary Data are available online at www.liebertpub.com/dna). Etiology of liver disease was divided by chronic hepatitis B, chronic hepatitis C, and others, including alcoholic liver disease, autoimmune hepatitis, Wilson’s disease, and other cryptogenic causes. Written informed consent was obtained from all study subjects. This study was approved by the Institutional Review Board at the Ajou University Hospital (GN3-07-400).
Array CGH
We used arrays of 4041 bacterial artificial chromosome clones (MACArray KARYO 4000 genome array; Macrogen, Inc., Seoul, Korea), including 1440 cancer-related genes, at a resolution of 1 Mb for scanning the entire genome for copy number changes. Each clone was printed in triplicate in a 12·12mm square area. The array platforms are described in detail in the Supplementary Materials and Methods. Images of fluorescence signals were acquired by a GenePix 4000V Microarray Scanner (Axon Instrument, Foster City, CA). Analysis of fluorescence signal intensities was performed using the array software package MAC Viewer (Macrogen, Inc.). Raw data for the 45 HCC samples are available from the Gene Expression Omnibus database (GSE65237).
Gene expression microarray profiling
Gene expression microarray analysis of the RNA from 31 tumors was performed using an Illumina HumanHT-12 v4 Expression BeadChip array (Illumina, Inc., San Diego, CA) according to the manufacturer’s instructions. The detailed information describing procedures of microarray analysis is available in the Supplementary Materials and Methods.
Statistical analysis
To analyze array CGH data, the automatic segmentation of DNA spots was applied and subtraction of background noise was performed, and thereafter, we calculated the total intensity and the intensity ratio of the two dyes for each spot. The resulting values were normalized based on intensity, and data were log2 transformed, and the mean log2 ratio (sample/reference) of the triplicate spots was calculated. A smoothing step was performed using locally weighted regression.
To verify genomic regions associated with copy number differences between normal and HCC samples, the chisquare test was performed with Monte Carlo simulation. The threshold for chromosomal gain and loss was defined as 0.001
Sharp, N.A., Luscombe, M.J., and Clemens, M.J. (1989). Regulation of c-fgr proto-oncogene expression in Burkitt’s lymphoma cells: effect of interferon treatment and relationship to EBV status and c-myc mRNA levels. Oncogene 4, 1043–1046.
Stranger, B.E., Forrest, M.S., Dunning, M., Ingle, C.E., Beazley, C., Thorne, N., et al. (2007). Relative impact of nucleotide and copy number variation on gene expression phenotypes. Science 315, 848–853.
Szczepanek, J., Pogorzala, M., Jarzab, M., Oczko-Wojciechowska, M., Kowalska, M., Tretyn, A., et al. (2012). Expression profiles of signal transduction genes in ex vivo drug-resistant pediatric acute lymphoblastic leukemia. Anticancer Res 32, 503–506.
Thomas,S.M.,andBrugge,J.S.(1997).Cellularfunctionsregulated by Src family kinases. Annu Rev Cell Dev Biol 13, 513–609.
Tomshine, J.C., Severson, S.R., Wigle, D.A., Sun, Z., Beleford, D.A., Shridhar, V., et al. (2009). Cell proliferation and epidermal growth factor signaling in non-small cell lung adenocarcinoma cell lines are dependent on Rin1. J Biol Chem 284, 26331–26339.
Yu, H.F., Zhao, G., Ge, Z.J., Wang, D.R., Chen, J., Zhang, Y., et al. (2012). High RIN1 expression is associated with poor prognosis in patients with gastric adenocarcinoma. Tumour Biol 33, 1557–1563.